Multi-Perspective Analysis of Household Carbon Dioxide Emissions from Direct Energy Consumption by the Methods of Logarithmic Mean Divisia Index and σ Convergence in Central China

Currently, household carbon dioxide (CO2) emissions (HCEs) as one of the leading sources of greenhouse gas (GHG) have drawn notable scholarly concern. Thus, here, taking six provinces in the period of 2000–2017 of Central China as a case, we analyzed the characteristics and the driving factors of HCEs from direct energy consumption and three perspectives: Central China as a whole, urban-rural differences, and inter-provincial comparison. The drivers of direct HCEs were analyzed by the Logarithmic Mean Divisia Index (LMDI). The σ convergence was adopted for analyzing the trend of inter-provincial differences on the HCEs. The key findings are as follows. First, all the direct HCEs from three perspectives had an obvious growth trend. The total direct HCEs grew from 9596.20 × 104 tonnes in 2000 to 30,318.35 × 104 tonnes in 2017, with an increase of 2.16 times. Electricity and coal use were the primary sources. The urban and rural increases of direct HCEs were up 2.57 times and 1.77 times, respectively. The urban-rural gap of direct HCEs narrowed first and then widened. The direct HCEs in the six provinces varied significantly, but the gap was narrowing. Second, as a whole the per capita consumption expenditure and energy demand were the main drivers to the increment of HCEs, with cumulative contribution rates of 118.19% and 59.90%. The energy price effect was mainly responsible for the mitigation of HCEs. Third, the similar drivers’ trend can also be seen from the perspective of inter-provincial comparison. However, from the perspective of urban and rural difference, the population urban-rural structure effect played a reverse influence on both urban and rural areas. Thus, raising the energy prices appropriately, upgrading the residents’ consumption to a sustainable pattern, controlling the growth of population size reasonably, and optimizing the household energy structure might effectively mitigate the growth of HCEs in Central China.


Introduction
Alleviating climate change is one of the key objectives to realize sustainable development. Human activity was found to have an important effect on global warming by the Intergovernmental Panel on Climate Change (IPCC) [1]. Therefore, reducing the emissions of man-made greenhouse gas (GHG) is one of the fundamental steps to address global climate change [2]. Moreover, the low-carbon economy has gradually become an important target worldwide [3,4]. China has been by far the biggest energy user and carbon series analysis. Using the LMDI model, Zha et al. [45], Shi et al. [40], and Zhou et al. [38] decomposed Chinese HCEs into carbon emissions coefficient, energy structure, energy intensity, income, population and so on. Yin et al. [46], Yang et al. [27], and Wang et al. [35] used a similar method to explore the influencing factors of HCEs in Beijing, Jiangxi, and Guangdong, respectively. Moreover, more and more scholars have used the STIRPAT model to analyze the impact factors of HCEs in recent years, such as Jin et al. [30], Wang et al. [47], Zhu et al. [48], Miao et al. [49], and Wang et al. [50]. The STIRPAT model is more extensible, and more socioeconomic factors can be incorporated. Therefore, the results from the model analysis are also more comprehensive [47].
In summary, the above empirical studies on HCEs are mainly conducted at the national level and single region or province of China as the case. The research perspectives focus mostly on a single analysis of urban-rural difference or regional comparison (Table 1). Few studies have been undertaken to integrate the HCEs from multiple perspectives for a particular region, particularly in underdeveloped or developing regions in China. Thus, in this paper, we take Central China as a study object. Figure 1 shows the location of six provinces in Central China. Central China is the second echelon of China's economic development, accounting for about 10.7% of the national land, and carrying approximately 26.51% of the country's population. Statistically, its GDP reached CNY 17.94 trillion in 2017. It plays a significant role in the geographical division of labor in China. Since the implementation of the "Central China Rising Strategy" in 2004, it has become an essential economic growth pole in China. However, meanwhile it can be easily seen that Central China has the enormous potential of increasing CO 2 emissions from household consumption. Therefore, it is crucial to study the HCEs in this region. So, we first demonstrated the characteristics and drivers of the HCEs from direct energy consumption and three perspectives: Central China as a whole, urban-rural differences, and inter-provincial comparison. The trend of inter-provincial differences on the HCEs was also analyzed. Finally, some specific countermeasures of emission reduction are proposed for this region's development based on the empirical results. The remainder of the paper is arranged as follows. The detailed methodology and data are in Section 2. The analysis and discussion of the empirical results are presented in Section 3. Section 4 concludes the results and proposes some reasonable countermeasures. Table 1. The summary of relevant studies on energy consumption and CO 2 emissions in the household sector.

Reviewed Paper
Location and Period Subject Scale Urban-Rural Differences

Regional Differences
Li et al. [19] China

Emission Coefficient Method
According to the IPCC [28], the direct HCEs were calculated by the consumption amounts of fuel and the corresponding carbon emission coefficients. The equation is: where C is the amount of direct HCEs (10 4 tonnes), i is the type of residents, i.e., urban residents and rural residents, j denotes the fuel type, Eij denotes the consumption of energy by resident i , and fj the carbon emission coefficient of j fuel type.

LMDI Model
In this paper, we used the LMDI model to explore the driving factors influencing the direct HCEs of Central China. This method can eliminate the residual term in the basic model and handle zero values in data [51,52]. While the basic principle of the LMDI model is the same, the formulas vary with the factors studied. The detailed formula can be expressed as the following extended identity:

Emission Coefficient Method
According to the IPCC [28], the direct HCEs were calculated by the consumption amounts of fuel and the corresponding carbon emission coefficients. The equation is: where C is the amount of direct HCEs (10 4 tonnes), i is the type of residents, i.e., urban residents and rural residents, j denotes the fuel type, E ij denotes the consumption of energy by resident i, and f j the carbon emission coefficient of j fuel type.

LMDI Model
In this paper, we used the LMDI model to explore the driving factors influencing the direct HCEs of Central China. This method can eliminate the residual term in the basic model and handle zero values in data [51,52]. While the basic principle of the LMDI model is the same, the formulas vary with the factors studied. The detailed formula can be expressed as the following extended identity: where the variables are defined in Table 2. Based on the above formula, we take C 0 and C t as the direct HCEs in the baseline period and target period. Thus, the change of total direct HCEs (C TOT ) can be written as: Notably, the CO 2 emission coefficients of various fuels are all considered constant when calculating the direct HCEs, which have no contribution to changes in CO 2 emissions. Thus, based on the additive LMDI method, we finally analyzed the following six effects: the energy structure effect (∆C ES ), energy price effect (∆C EP ), energy demand effect (∆C ED ), per capita consumption expenditure effect (∆C PC ), population urban-rural structure effect (∆C PS ), and population size effect (∆C P ).
Each sub-effect can be expressed as:

Classical Convergence Regression Method-σ Convergence
The trend of inter-provincial differences in direct HCEs across the six provinces can be determined using the classical convergence regression method. The σ convergence is mainly tested by the coefficient of variation, which is expressed as: where CV t denotes the coefficient of variation in year t and C k,t denotes the direct HCEs of the k province in year t. If CV t keeps decreasing, it means that there is σ convergence in Sustainability 2021, 13, 9285 6 of 25 the direct HCEs in the six provinces, and the difference of direct HCEs in each province is gradually decreasing.

Description of the Data
The data on household energy consumption for urban and rural areas in the six provinces are found in the China Energy Statistical Yearbook (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018) [53]. The energy sources involved in measuring the direct HCEs are raw coal, washed coal, other washed coal, briquettes, coke, coke oven gas, blast furnace gas, converter gas, other gas, gasoline, kerosene, diesel, fuel oil, lubricants, petroleum asphalt, liquefied petroleum gas (LPG), refinery dry gas, other petroleum products, natural gas, liquefied natural gas (LNG), electricity, heat, and other types. It should be noted that heat and electricity are the secondary energy sources with no direct use of fuel, while a large amount of fuel is consumed in the production process. Thus, in this paper, electricity and heat are also considered. The emission coefficients for different types of primary energy can be calculated using the method recommended by the IPCC (2006), as listed in Table A1. Given the complexity of calculating emission factors from electricity and heat, we used the statistical data on average CO 2 emissions per unit of electricity supply in the Guidelines for the Preparation of Provincial Greenhouse Gas Inventories (for Trial Implementation) [54]. The emission factors for electricity are shown in Table A2. The emission coefficient of heat adopted 0.11 t-CO 2 /GJ according to the reference of Jia et al. [55]. Relevant data of influencing factors such as consumption expenditure and population size in the six provinces are from Statistical Yearbooks (2001-2018) [56][57][58][59][60][61]. All the economic data are comparable in that they are converted into 2000 standard prices. The CO 2 emission coefficient for standard coal is 2.4567 [54]. The conversion factor of standard coal for various energy types is from the China Energy Statistical Yearbook (Table A1).  Table 3). The per capita direct HCEs rose from 0.27 tonnes to 0.77 tonnes, with an average annual increase amount and rate of 0.03 tonnes and 6.32%, respectively. All energy types were divided into five categories, and the electricity and coal consumption dominated the direct HCEs. For example, the cumulative share of CO 2 emissions from the electricity during 2000-2017, accounted for 46.55% of total HCEs in Central China (Figure 2b).  Table 3). The per capita direct HCEs rose from 0.27 tonnes to 0.77 tonnes, with an average annual increase amount and rate of 0.03 tonnes and 6.32%, respectively. All energy types were divided into five categories, and the electricity and coal consumption dominated the direct HCEs. For example, the cumulative share of CO2 emissions from the electricity during 2000−2017, accounted for 46.55% of total HCEs in Central China (Figure 2b).
(a) (b)  The structure for direct HCEs has varied significantly in the 2000−2017 years ( Figure  2a). The coal-related emissions dropped remarkably from 6423.21 × 10 4 tonnes to 4875.82 × 10 4 tonnes, and their share decreased at an average annual rate of −1.61% from 66.93% in 2000 to 16.08% in 2017. This trend can be explained by three reasons. Firstly, the utilization rate of coal was improved due to the revolution of coal engineering by technological progress. Secondly, coal-to-gas conversion has been implemented in several locations by China's policy against dry fog. Residential heating has been converted to a centralized  16.08% in 2017. This trend can be explained by three reasons. Firstly, the utilization rate of coal was improved due to the revolution of coal engineering by technological progress. Secondly, coal-to-gas conversion has been implemented in several locations by China's policy against dry fog. Residential heating has been converted to a centralized type in recent years. Thirdly, the global financial crisis caused a rapid rise in coal prices. On the contrary, the CO 2 emissions generated by four other energy use were all up. Significantly, the CO 2 emissions of oil grew fastest, followed by electricity. Their average annual growth rates were 16.44% and 11.82%, respectively. In 2017, the shares of CO 2 emissions from oil and electricity were up to 26.25% and 55.53%. Meanwhile, the CO 2 emissions induced by oil have surpassed those from coal, making it the second-largest energy type. Oil, like diesel, was mainly used for transportation. Since people's demand for transportation has escalated with household income, the number of people purchasing motorcycles and family cars increased. Statistically, the average number of motorcycles and family cars per 100 households owned by residents of Central China was 43.35 and 23.08 in 2017, respectively, which were about 2.75 times and 128.00 times that of 2000 (Table 4). In addition, the growth in household appliances (especially those that consume large amounts of electricity, such as air conditioners) has increased the CO 2 emissions from electricity. The average number of air conditioners per 100 households owned by residents of Central China increased by 616.48% from 13.29 units in 2000 to 95.22 units in 2017. The share of CO 2 emissions caused by natural gas has increased over the past few years, while its overall proportion was relatively low at just 4.85% in 2017. People would prefer cleaner and more convenient natural gas in their energy choices with economic growth and improved living standards. However, the economy of the study region is slightly lower than that of Eastern China, resulting in relatively low use of natural gas. The CO 2 emissions of heat also presented an increased tendency (212.64 × 10 4 tonnes in 2000 and 2157.89 × 10 4 tonnes in 2017), the share of which increased from 2.22% in 2000 to 7.12% in 2017. In general, the energy use structure has been progressively improved and decarbonized, and the share of clean energy and low-carbon energy use has risen. However, given that heat and electricity are produced with coal as the raw material, the energy structure has not been completely altered.

Urban-Rural Differences
It can be seen that the direct HCEs of both urban and rural households have risen continuously ( Figure 3). More CO 2 emissions were caused by urban household in most years. The urban and rural direct HCEs significantly increased by 2.57 times (from 4679.08 × 10 4 tonnes to 16,698.56 × 10 4 tonnes) and 1.77 times (from 4917.12 × 10 4 tonnes to 13,619.79 × 10 4 tonnes), respectively. The corresponding annual average growth rates were 7.77% and 6.18%. They accounted for approximately 20% and 30% of national urban and rural direct HCEs, respectively (Table 3). For the entire study period, the growth amount and rate in urban areas were apparently higher than that of rural. This can be divided into two stages. From 2000 to 2006, rural direct HCEs were higher than those in urban areas. The average annual growth rate of direct HCEs in rural areas was 6.48%, which was lower than that of 7.16% in urban. The urban-rural ratio of direct HCEs increased to 0.99 from 0.95, gradually approaching 1. Thus, the gap between direct HCEs for urban and rural regions was gradually narrowing. After 2006, the direct HCEs in urban areas began to be higher than those of rural areas, and the urban-rural ratio rose from 0.99 to 1.23, with the gap gradually widening again. The result was consistent with previous studies, such as Zhang at al. [62]. This could be caused by the rapid urbanization. The commercial energy use has increased due to urban population growth, while rural areas are still dominated by non-commercial energy after 2006 [29].
Sustainability 2021, 13, x FOR PEER REVIEW 9 of 28 0.95, gradually approaching 1. Thus, the gap between direct HCEs for urban and rural regions was gradually narrowing. After 2006, the direct HCEs in urban areas began to be higher than those of rural areas, and the urban-rural ratio rose from 0.99 to 1.23, with the gap gradually widening again. The result was consistent with previous studies, such as Zhang at al. [62]. This could be caused by the rapid urbanization. The commercial energy use has increased due to urban population growth, while rural areas are still dominated by non-commercial energy after 2006 [29]. There were also many significant differences in the structure of direct HCEs between urban and rural regions. The rural direct HCEs caused by coal consumption were consid-  There were also many significant differences in the structure of direct HCEs between urban and rural regions. The rural direct HCEs caused by coal consumption were considerably higher than those of urban (Figure 4b), whereas the other four types of energy use were opposite (Figure 4a). A typical dichotomy existed between urban and rural areas in China [29]. The urban-rural differences in the direct HCEs structure were likely caused by substantial variability in income level, energy supply, and energy use habits of urban and rural residents. At present, centralized heating and gas supply have not been taken yet in most rural areas of Central China. The residents mainly use coal for heating and cooking, resulting in heavy coal use. The economic conditions, quality of life, and the ownership of vehicles and household appliances (washing machines, refrigerators, and air conditioners) in urban areas were higher than in rural areas (Table 5), which resulted in greater consumption of oil and electricity in urban areas than in rural. As shown in Table 5, the average number of family cars owned by urban households (per 100 households) was as high as 31.50, while the number was only 14.40 in rural areas in 2017. Natural gas was mainly used for heating and cooking, and indoor heating was centralized for urban dwellers. Urban energy was supplied centrally, which had a specific intensive and scale effect compared to rural areas.

Inter-Provincial Comparison
In terms of inter-provincial comparison, there was an evident disparity in direct HCEs among the six provinces. Henan Province is the most populous province contributing the largest HCEs during 2000-2017. It contributed a 24.71% to the total HCEs of Central China, followed by Shanxi (18.52%), Hubei (17.12%), Hunan (15.98%), Anhui (15.11%), and Jiangxi (8.56%). Remarkably, the direct HCEs from the six provinces all displayed a fluctuating upward trend during 2000-2017 (Figure 5a). The lowest level of direct HCEs was in Jiangxi, while the overall tendency was on the rise, with an increase of 288.61% and an average annual growth rate of 8.31%. On the contrary, the direct HCEs of Henan Province were always the highest during the same period, whereas the growth amount and rate were both lower than that of Jiangxi, with an increase of about 175.42% and an average annual growth rate of 6.14%.  To understand whether the difference of direct HCEs among the six provinces is narrowing or widening, we adopted the σ convergence test in neoclassical theory to describe it. The coefficient of variation of direct HCEs was used to test. If the coefficient of variation shows a downward trend, there is convergence, indicating that the difference of direct HCEs in the six provinces is progressively decreasing. Based on Equation (10) It can also be easily seen that the direct HCEs gap was decreasing in the six provinces from an overall perspective.
The cumulative shares of CO2 emissions from five energy types in each province were calculated ( Figure 6). Then, the emission structure of each province was compared and analyzed. First, Shanxi Province took the largest share of CO2 from coal use at 45.14% in comparison with other provinces, followed by Henan (36.92%), Hubei (34.59%), Hunan (31.92%), and Anhui (27.71%). The share of CO2 emissions from coal use in Jiangxi was the lowest, at 26.86%. The coal resources of Shanxi Province were the richest compared to the other five provinces, with more production and usage, so the share of coal The HCEs of the remaining four provinces (Hunan, Anhui, Shanxi, and Hubei) were at a medium level during 2000-2017, which can be roughly analyzed in two stages (2000-2005 and 2005-2017). From 2000 to 2005, Hunan Province had the lowest direct HCEs, while it had a speedy growth trend with a growth rate much higher than the other three provinces. The average annual growth rate was 14.9% and the total increase amount was about 100.3%. The average annual growth rate of Anhui (Shanxi) was 8.64% (3.87%), with the total increase amounts of about 51.32% (20.9%). Hubei's HCEs had the lowest average annual growth rate of only 1.98%. During 2005-2017, Hunan Province had the highest growth rate of direct HCEs among the four provinces, with an average annual growth rate of 7.9%. Beginning in 2015, the direct HCEs of Hunan reached the highest level among the four provinces gradually and significantly, reaching 5489.00 × 10 4 tonnes in 2017. Over this period, there was no significant disparity in the growth rate of direct HCEs for the four provinces. The average annual growth rates of other three provinces (Shanxi, Hubei, and Anhui) were 6.88%, 7.77%, and 6.77%, respectively. Their direct HCEs in 2017 were 5145.26 × 10 4 tonnes, 4932.79 × 10 4 tonnes, and 4797.57 × 10 4 tonnes.
To understand whether the difference of direct HCEs among the six provinces is narrowing or widening, we adopted the σ convergence test in neoclassical theory to describe it. The coefficient of variation of direct HCEs was used to test. If the coefficient of variation shows a downward trend, there is convergence, indicating that the difference of direct HCEs in the six provinces is progressively decreasing. Based on Equation (10), the coefficient of variation of direct HCEs in the six provinces was calculated. The result indicated that the coefficient of variation exhibited an interval increase and decrease from 2000-2017 (Figure 5b). From 2000 to 2003, the coefficient of variation of direct HCEs presented a linearly decreasing trend. The σ convergence was obvious over this period. The difference in direct HCEs was steadily declining among the provinces. From 2003 to 2012, the trend in the coefficient of variation was fluctuating upwards, indicating that there was no σ convergence in this period. From 2012 to 2017, it declined first and rebounded somewhat in 2017, which is a dispersion stage. The disparity in direct HCEs increased among the provinces. Overall, the coefficient of variation of direct HCEs decreased from 0.35 in 2000 to 0.25 in 2017, indicating that there was σ convergence of direct HCEs in the six provinces. It can also be easily seen that the direct HCEs gap was decreasing in the six provinces from an overall perspective.
The cumulative shares of CO 2 emissions from five energy types in each province were calculated ( Figure 6). Then, the emission structure of each province was compared and analyzed. First, Shanxi Province took the largest share of CO 2 from coal use at 45.14% in comparison with other provinces, followed by Henan (36.92%), Hubei (34.59%), Hunan (31.92%), and Anhui (27.71%). The share of CO 2 emissions from coal use in Jiangxi was the lowest, at 26.86%. The coal resources of Shanxi Province were the richest compared to the other five provinces, with more production and usage, so the share of coal consumption and CO 2 emissions was the highest. The largest share of CO 2 emissions induced by oil was in Hubei, and the smallest share was in Shanxi, with 16.17% and 3.24%, respectively. The share of CO 2 emissions caused by natural gas was relatively small in all six provinces, ranging from 1.72% to 4.21%. The share of Anhui was the highest and Hunan Province was the lowest. The CO 2 emissions generated by electricity consumption represented a large proportion, with Jiangxi, Hunan, and Anhui all exceeding 50%, while Shanxi represented the least, at only 34.38%. The share of CO 2 emissions from heat use in Shanxi was high, at 14.49%. Thus, it can be seen that Shanxi Province had a poor low-carbon energy structure.

Overall Analysis
The complete decomposition results are presented in Figure 7 and Table A3. The cumulative contribution of each driver was calculated from 2000 to 2017 for the ease of analysis (Figure 8a). During the whole study, the change of total direct HCEs was 20,722.15 × 10 4 tonnes. Generally, the energy structure effect, energy demand effect, per capita consumption expenditure effect, population urban-rural structure effect, and population size effect positively affected the growth of direct HCEs. On the contrary, the energy price effect showed a negative impact, which decreased the direct HCEs. Next, the contribution of each driving factor at four different stages (2000-2005, 2005-2010, 2010-2015, and 2015-2017) is drawn out in Figure 8b.

Overall Analysis
The complete decomposition results are presented in Figure 7 and Table 3. The cumulative contribution of each driver was calculated from 2000 to 2017 for the ease of analysis (Figure 8a). During the whole study, the change of total direct HCEs was 20,722.15 × 10 4 tonnes. Generally, the energy structure effect, energy demand effect, per capita consumption expenditure effect, population urban-rural structure effect, and popu-lation size effect positively affected the growth of direct HCEs. On the contrary, the energy price effect showed a negative impact, which decreased the direct HCEs. Next, the contribution of each driving factor at four different stages (2000-2005, 2005-

Energy Price Effect
The energy price effect was mainly responsible for mitigating direct HCEs in Central China (Figure 7 and Figure 8a), which was consistent with the previous studies of Yang et al. and Wang et al. [27,35]. However, it was also contrary to the results of a few studies, such as Zhao et al. [63].  (Table A3 and Figure 8b). The corresponding average annual contribution rates in the four stages were 0.37%, −5.75%, −12.61%, and −6.24%, respectively (Table 6). This mitigation effect may be caused by the increase in the average energy prices during

Energy Price Effect
The energy price effect was mainly responsible for mitigating direct HCEs in Central China (Figure 7 and Figure 8a), which was consistent with the previous studies of Yang et al. and Wang et al. [27,35]. However, it was also contrary to the results of a few studies, such as Zhao et al. [63].  (Table A3 and Figure 8b). The corresponding average annual contribution rates in the four stages were 0.37%, −5.75%, −12.61%, and −6.24%, respectively (Table 6). This mitigation effect may be caused by the increase in the average energy prices during

Energy Price Effect
The energy price effect was mainly responsible for mitigating direct HCEs in Central China (Figures 7 and 8a), which was consistent with the previous studies of Yang et al. and Wang et al. [27,35]. However, it was also contrary to the results of a few studies, such as Zhao et al. [63].  (Table 3 and Figure 8b). The corresponding average annual contribution rates in the four stages were 0.37%, −5.75%, −12.61%, and −6.24%, respectively (Table 6). This mitigation effect may be caused by the increase in the average energy prices during the study period. The Chinese government has launched price deregulation in the energy sector, which has led to an overall increase in energy prices. From 2000 to 2017, the average energy prices in Central China increased to 1.02 × 10 4 CNY/tonne of standard coal equivalent (tce) from 0.48 × 10 4 CNY/tce, with an average annual growth rate of 4.55% (Figure 9a). The average annual growth rates of energy prices in the four phases were 4.07%, 0.82%, 5.65%, and 2.26%, respectively. Obviously, the energy price effect had the strongest impact on the decrease of direct HCEs in 2010-2015.

Urban-Rural Differences
The decomposition results of direct HCEs from urban and rural areas during 2000-2017 are presented in Table A4 and Table A5. The same drivers affected urban and rural areas differently. As a whole, the energy structure effect, energy demand effect, per capita consumption expenditure effect, and population effect all contributed to the growth of direct HCEs in both urban and rural areas during 2000-2017 ( Figure 10). The effects' contribution of the energy structures and per capita consumption expenditure contributed more to rural compared with urban areas. In 2017, these drivers increased direct HCEs by

Per Capita Consumption Expenditure Effect and Energy Demand Effect
The per capita consumption expenditure may, to some extent, reflect the living standard of a region. It represents the residential energy consumption level with the energy demand together. The per capita consumption expenditure effect played the most important role in promoting the growth of direct HCEs (Figures 7 and 8a). It contributed a total of 24,490.81 × 10 4 tonnes CO 2 over the study period and accounted for 118.19% of total change of direct HCEs. Meanwhile, the energy demand was another major contributor for the increment of direct HCEs. It contributed 12,413.03 × 10 4 tonnes CO 2 and the contribution rate was 59.90%. Concretely, in the four stages, the corresponding contribution values (rates) were 3757.91 × 10 4 tonnes (18.13%), 6130.93 × 10 4 tonnes (29.59%), 10,926.11 × 10 4 tonnes (52.73%), and 3675.86 × 10 4 tonnes (17.74%) for the per capita consumption expenditure effect, and −1472.68 × 10 4 tonnes (−7.11%), 3051.96 × 10 4 tonnes (14.73%), 9511.01 ×10 4 tonnes (45.90%), and 1322.74 × 10 4 tonnes (6.38%) for the energy demand effect (Figure 8b and Table 3). The average annual contribution rates of the per capita consumption expenditure effect (energy demand effect) in the four stages were 3.63% (−1.42%), 5.92% (2.95%), 10.55% (9.18%), and 8.87% (3.19%), respectively ( Table 6). It can be seen that the driving effect of per capita consumption expenditure and energy demand on the total direct HCEs was increasing. However, their driving role declined in 2015−2017. The continued increase in residence expenditure has promoted the continued growth of HCEs directly. During 2000-2017, the per capita residence expenditure in Central China rose from CNY 290.11 to CNY 2064.30, an increase of 611.57%, with an average annual growth rate of 12.24% (Figure 9c). The average annual growth rates were 5.32%, 9.77%, 18.42%, and 10.76%, respectively in the four stages. The overall increase in household consumption level was also an essential reason for the rapid growth of direct HCEs in this region. Throughout the study period, the per capita consumption expenditure was up from CNY 2309.46 to CNY 10,087.93, an increase of 336.81%, with an average annual growth rate of 9.06% (Figure 9b). The average yearly growth rates in the four phases were 8.62%, 6.96%, 7.28%, and 3.65%, respectively. Therefore, people had the capacity to purchase more energy-intensive appliances and transportation tools, resulting in greater energy use. The number of air conditioners, refrigerators, and family cars owned per 100 households at the 2017 year-end were 95.22, 94.32, and 23.08, compared to 13.29, 43.44, and 0.18 in 2000, with an increase of 616.48%, 117.13%, and 12,722.22%. Hence, it is vital to reduce the direct HCEs by limiting the growth of transportation tools and high energy-consuming appliances and improving energy use efficiency.
There are two reasons why the population urban-rural structure showed a positive effect. One was that the per capita direct HCEs in urban areas were all higher than those in rural; the other was that the proportion of urban population continued to rise, and the urbanization rate kept increasing, owing to the continuous urbanization process. The urban population ratio rose by 63.80% from 28.22% in 2000 to 46.23% in 2017, with an average annual growth rate of 2.95% (Figure 9d). The average annual growth rates were 3.73%, 3.31%, 1.41%, and 3.97% for each of the four stages, respectively.

Energy Structure Effect
The energy structure effect played a minor role in increasing the direct HCEs, which contributed 1743.56 × 10 4 tonnes CO 2 over this 17-year period and the contribution rate was 8.41%. Specifically, the contribution values (rates) at each of the four stages were 603.48 × 10 4 tonnes (2.91%), 935.23 × 10 4 tonnes (4.51%), −409.25 × 10 4 tonnes (−1.97%), and 614.10 × 10 4 tonnes (2.96%), respectively (Figure 8b and Table 3). The corresponding average annual contribution rates of the energy structure effect were 0.58%, 0.90%, −0.39%, and 1.48%, respectively ( Table 6). It should be particularly noted that the growth of direct HCEs was inhibited during 2010-2015. The reason could be that the energy structure was being optimized in Central China at this period, which reduced CO 2 emissions. As mentioned above, it was clear that CO 2 emissions from coal use were gradually falling, while emissions from oil, natural gas, heat, and electricity were rising. China is constrained by coal-based resource endowments and the production of electricity and heat are also based on coal as the primary raw material. Consequently, coal is used in great quantities. Therefore, the government should strengthen the encouragement and support for developing non-fossil fuel energy sources such as wind energy, nuclear power, biomass, solar energy, and other clean energy. Natural gas is also a relatively low-carbon emitting energy of fossil fuels. At the same time, the consumption share of natural gas has been low due to the relatively underdeveloped economic conditions in Central China. Therefore, the natural gas industry should be strongly supported in this region, especially in rural areas, so that natural gas can be widely used. It is also an issue worthy of attention in the decarbonization of household consumption.

Urban-Rural Differences
The decomposition results of direct HCEs from urban and rural areas during 2000-2017 are presented in Tables 4 and 5. The same drivers affected urban and rural areas differently. As a whole, the energy structure effect, energy demand effect, per capita consumption expenditure effect, and population effect all contributed to the growth of direct HCEs in both urban and rural areas during 2000-2017 ( Figure 10). The effects' contribution of the energy structures and per capita consumption expenditure contributed more to rural compared with urban areas. In 2017, these drivers increased direct HCEs by 72.07% and 54.53% in rural and urban areas. Conversely, the effects of energy demand and population in urban areas were greater than that in rural, which increased by 45.42% and 23.88% of total CO 2 emissions. The energy price effect was negative for urban and rural direct HCEs, with cumulative contribution rates of −61.70% and −40.74%, respectively.
The differences in drivers between urban and rural areas can be explained by different change rates of various influencing factors. For example, the difference in the magnitude of the energy price effect can be explained by the variation in the average energy prices for urban and rural areas. The energy prices rose from 0.44 × 10 4 CNY/tce and 0.52 × 10 4 CNY/tce in 2000 to 1.11 × 10 4 CNY/tce and 0.93 × 10 4 CNY/tce in 2017 for urban and rural regions, respectively. In terms of the growth rate, compared to 2000, it was 153.95% for urban areas and 78.69% for rural (Figure 11a). The change rates of per capita consumption expenditure for urban and rural areas were 228.31% and 344.13%, respectively (Figure 11b). The growth rate of urban energy demand was higher than that of rural areas during the study period, as depicted in Figure 11c. The energy demand of urban and rural areas rose from 11.59% and 13.54% to 19.39% and 22.26%, with the growth rates of 67.29% and 64.44%, respectively (Figure 11c). The population urban-rural structure effect promoted the growth of direct HCEs in urban areas, whereas the opposite was true for rural. The contribution rates for urban and rural areas were 34.05% and −31.46%, respectively. By 2017, the ratio of urban and rural population increased by 63.8% and −25.1% than that in 2000 (Figure 11d). and 64.44%, respectively (Figure 11c). The population urban-rural structure effect promoted the growth of direct HCEs in urban areas, whereas the opposite was true for rural. The contribution rates for urban and rural areas were 34.05% and −31.46%, respectively. By 2017, the ratio of urban and rural population increased by 63.8% and −25.1% than that in 2000 (Figure 11d). In short, the two drivers with the greatest degree of influence on direct HCEs in rural areas were the effects of per capita consumption expenditure and energy demand, cumulatively contributing 85.14% of total HCEs. The population urban-rural structure effect and energy price effect reduced 53.96% of total direct HCEs. In urban areas, the two factors that contributed most to direct HCEs were also the per capita consumption expenditure and energy demand, contributing about 92.95% of direct HCEs. The change of energy price in urban areas mitigated the direct HCEs by 61.70%.

Inter-Provincial Comparison
The cumulative contributions of various drivers to the HCEs for each province are exhibited in Figure 12, and the inter-provincial differences are noticeable. In short, the two drivers with the greatest degree of influence on direct HCEs in rural areas were the effects of per capita consumption expenditure and energy demand, cumulatively contributing 85.14% of total HCEs. The population urban-rural structure effect and energy price effect reduced 53.96% of total direct HCEs. In urban areas, the two factors that contributed most to direct HCEs were also the per capita consumption expenditure and energy demand, contributing about 92.95% of direct HCEs. The change of energy price in urban areas mitigated the direct HCEs by 61.70%.

Inter-Provincial Comparison
The cumulative contributions of various drivers to the HCEs for each province are exhibited in Figure 12, and the inter-provincial differences are noticeable. The energy price effect mitigated the growth of direct HCEs in all six provinces from 2000−2017. The absolute value of cumulative contribution of Henan Province was the largest (25.70%), followed by Shanxi (24.21%). The variation of energy prices in Hunan mitigated the smallest direct HCEs (8.11%). This could be explained by the prominent gap of the change rate in average energy prices among provinces. The energy prices of Hunan, Anhui, Jiangxi, Hubei, Henan, and Shanxi increased by 21.86%, 117.46%, 125.17%, 155.55%, 197.57%, and 272.10%, respectively (Figure 13a). The more energy prices rise, the stronger the contribution to carbon reduction will be. Therefore, Hunan Province had the largest flexibility to increase energy prices, while the opposite was true for Henan Province. It indicated that Hunan has a greater potential to reduce the direct HCEs by raising energy prices relative to Henan.
The per capita consumption expenditure effect boosted the growth of direct HCEs in all provinces. The largest contribution to direct HCEs was found in Henan, with a contribution rate of 29.44%. Conversely, the change of the per capita consumption expenditure in Jiangxi contributed the least direct HCEs (9.72%). Shanxi, Hubei, Anhui, and Hunan, whose contribution rates were 21.26%, 19.75%, 19.55%, and 18.46%, respectively. From 2000-2017, the per capita consumption expenditure in Shanxi Province and Jiangxi Province increased by 390.89% and 361.05%, while Hunan Province increased by only 251.65% (Figure 13b). It can be seen that the difference in the per capita consumption expenditure effect was mainly caused by the change of the per capita consumption expenditure for each province.
The population effect likewise contributed the most strongly to direct HCEs in Henan Province, followed by Shanxi, with cumulative contribution rates of 2.88% and 2.12%, respectively. The pulling effect on direct HCEs was more remarkable than in the other four provinces (Hunan, Jiangxi, Hubei, and Anhui). The population effect made the smallest The energy price effect mitigated the growth of direct HCEs in all six provinces from 2000-2017. The absolute value of cumulative contribution of Henan Province was the largest (25.70%), followed by Shanxi (24.21%). The variation of energy prices in Hunan mitigated the smallest direct HCEs (8.11%). This could be explained by the prominent gap of the change rate in average energy prices among provinces. The energy prices of Hunan, Anhui, Jiangxi, Hubei, Henan, and Shanxi increased by 21.86%, 117.46%, 125.17%, 155.55%, 197.57%, and 272.10%, respectively (Figure 13a). The more energy prices rise, the stronger the contribution to carbon reduction will be. Therefore, Hunan Province had the largest flexibility to increase energy prices, while the opposite was true for Henan Province. It indicated that Hunan has a greater potential to reduce the direct HCEs by raising energy prices relative to Henan.
The per capita consumption expenditure effect boosted the growth of direct HCEs in all provinces. The largest contribution to direct HCEs was found in Henan, with a contribution rate of 29.44%. Conversely, the change of the per capita consumption expenditure in Jiangxi contributed the least direct HCEs (9.72%). Shanxi, Hubei, Anhui, and Hunan, whose contribution rates were 21.26%, 19.75%, 19.55%, and 18.46%, respectively. From 2000-2017, the per capita consumption expenditure in Shanxi Province and Jiangxi Province increased by 390.89% and 361.05%, while Hunan Province increased by only 251.65% (Figure 13b). It can be seen that the difference in the per capita consumption expenditure effect was mainly caused by the change of the per capita consumption expenditure for each province.

Main Conclusions
Taking Central China in the 2000-2017 years as an example, we analyzed the characteristics and driving factors of the direct HCEs from three perspectives. The urban-rural differences and inter-provincial comparison were highlighted. The conclusions are set out below. The population effect likewise contributed the most strongly to direct HCEs in Henan Province, followed by Shanxi, with cumulative contribution rates of 2.88% and 2.12%, respectively. The pulling effect on direct HCEs was more remarkable than in the other four provinces (Hunan, Jiangxi, Hubei, and Anhui). The population effect made the smallest contribution to Jiangxi, by only 0.76%. From 2000 to 2017, the population of Henan, Shanxi and Jiangxi provinces was increasing year by year. The largest growth rate of population was 14.39% in Henan, and the growth rates of population of Shanxi, Anhui, Hunan, Hubei, and Jiangxi were 14.00%, 12.44%, 11.18%, 4.53%, and 11.41%, respectively (Figure 13c), indicating that the gap of population growth was responsible for the difference of the population effect in the six provinces.
The energy structure effect was only negative in Anhui Province, with a cumulative contribution rate of −1.15% during 2000-2017. This can be explained by the change of energy structure in Anhui reduced the direct HCEs overall. Conversely, the energy structure effect in Shanxi contributed the highest direct HCEs (3.94%), indicating that the shift in Shanxi's energy structure was most unfavorable to the reduction of direct HCEs. Therefore, it is more imperative for Shanxi to adjust the energy structure effectively by reducing the use of fossil energy such as coal and oil and accelerating the development of non-fossil energy sources such as wind energy and nuclear power.
The population urban-rural structure effect only showed a negative effect on direct HCEs in Hunan Province. The contribution rate of the population urban-rural structure effect varied significantly between provinces from 2000-2017. The aggregate contribution of the population urban-rural structure effect in Anhui Province was the highest at 2.64%, followed by Henan Province (2.39%), and Hunan Province had the lowest cumulative contribution of −1.15%. It can be derived that the change of population urban-rural structure in Hunan Province was generally beneficial to mitigating the direct HCEs. In the survey period, the share of urban population in Henan province increased from 23.20% to 50.16%, with an increase of 116.23%, while the proportion of urban population in Hunan province increased from 29.75% to 33.54%, with a growth rate of only 12.73% (Figure 13d). Thus, it can be seen that the inter-provincial differences in the contribution of population urban-rural structure effect were mainly caused by the different change rates of the urban population share in each province.
The energy demand effect showed the strongest impact on the direct HCEs growth in Shanxi Province (11.51%), and the smallest impact on the increment of direct HCEs in Jiangxi Province (6.97%). The effect of energy demand in Hubei, Henan, Hunan, and Anhui contributed 10.74%, 10.52%, 10.36%, and 9.81% to total change of direct HCEs, respectively ( Figure 12). Energy demand in Shanxi Province increased from 10.01% in 2000 to 18.39% in 2017, with a growth rate of 83.65%. The change rates of energy demand in Hubei, Henan, Hunan, Anhui, and Jiangxi were 49.89%, 33.04%, 60.71%, 86.15%, and 72.39%, respectively (Figure 13e). This difference was mainly due to the different extent of variation in the share of residential energy consumption between provinces.

Main Conclusions
Taking Central China in the 2000-2017 years as an example, we analyzed the characteristics and driving factors of the direct HCEs from three perspectives. The urban-rural differences and inter-provincial comparison were highlighted. The conclusions are set out below.
From 2000 to 2017, total direct HCEs increased at an average annual growth of 7.00% from 9596.20 × 10 4 tonnes to 30,318.35 × 10 4 tonnes. Electricity and coal consumption were the primary sources. The direct HCEs in both urban and rural rose continuously. The growth amount and rate of urban direct HCEs were all higher than those from rural areas. The urban-rural gap of direct HCEs narrowed during 2000-2006 and then widened after 2006. There were also significant differences in the structure of direct HCEs between urban and rural areas. The share of CO 2 emissions generated by coal consumption in rural areas was higher than that in urban, whereas CO 2 emissions from oil, natural gas, heat, and electricity use were the opposite. From the perspective of inter-provincial comparison, the direct HCEs in the six provinces were all on the rise, while the growth rate and amount differed significantly. Among these, Henan, Shanxi, Hubei, Hunan, Anhui, and Jiangxi occupied 24.71%, 18.52%, 17.12%, 15.98%, 15.11%, and 8.56%, respectively, in the order of cumulative value. The difference in the amount of direct HCEs among provinces was decreasing.
Overall, the effects' contribution of energy structure, energy price, energy demand, per capita consumption expenditure, population urban-rural structure and population were 8.41%, −102.44%, 59.90%, 118.19%, 6.54%, and 9.40%, respectively. The results indicated that the growth in per capita consumption expenditure was the most prominent driver behind the increase of direct HCEs in Central China, followed by energy demand. The effects of population size, energy structure and population urban-rural structure had less impact on the growth of direct HCEs. However, energy price presented the most obvious mitigation effect. From the perspective of urban-rural differences, the effects of per capita consumption expenditure and energy structure had a greater impact on direct HCEs in rural areas than in urban among the contributing factors, whereas the energy demand effect and the population size effect were the opposite. The energy price effect was the greater disincentive on urban areas than on rural. The effect of population urban-rural structure promoted the growth of direct HCEs in urban areas, whereas the opposite was true for rural. There were also some variabilities in the extent to which the same driving factor affected different provinces.

Discussion
It would be meaningful to compare the results of this paper with previous studies. First of all, the direct HCEs had an obvious growth trend in Central China, with an average annual growth of 7.00%. On the other hand, the direct HCEs of China increased by 51,883 × 10 4 tonnes during 1996−2012 [19], and the direct HCEs in China grew at an average annual rate of 7% from 2000-2015 [40]. Therefore, it can be seen that the growth rate of total direct HCEs in Central China is similar to that of the whole of China. Additionally, coal and electricity consumption were the main sources of HCEs. The cumulative share of CO 2 emissions from electricity and coal during 2000-2017 accounted for 46.55% and 34.99% of the total HCEs in Central China. Nationally, the total share of CO 2 emissions from coal and electricity ranged from 45.7% to 61.4% [45]. Moreover, the direct HCEs and growth rate in urban areas were higher than those in rural areas, and the urban-rural differences were significant, which was in line with Li et al. [19] and Feng et al. [33]. From a single year, the direct HCEs in Central China were 26,535.65 × 10 4 tonnes in 2015, accounting for approximately 22.85% of the national direct HCEs (11,6120 × 10 4 tonnes) [40].
The LMDI decomposition analysis was then applied to investigate the factors that might affect the changes of the direct HCEs. The decomposition results in Central China were similar to those of Zha et al. in the whole of China [45]. They both indicated that the energy price (energy intensity) and per capita consumption expenditure (income) effects contributed most to the decline and the increase of CO 2 emissions, respectively. Other factors had less impact on the growth of CO 2 emissions. However, our decomposition results differed from those in Guangdong Province [35]. The first promoting factor to HCEs in Guangdong were population size (1996−2004) and energy use level (2005−2016). The carbon emission coefficient was the main inhibiting factor. This fact could be explained by the differences in the level of economic development and resource endowments between regions.
In addition, there are some limitations in this paper. Only six driving factors affecting the direct HCEs in Central China were explored, specifically energy structure, per capita consumption expenditure, energy price, energy demand, population urban-rural structure, and population size. However, direct HCEs may be influenced by the number of households [48], temperature, Engel coefficient [30], demographic structure [48], and other socioeconomic factors. Thus, in our future work, these factors to the direct HCEs should be given greater attention.

Suggestions
Based on these results, some particular countermeasures and suggestions for sustainable household energy consumption of Central China are put forward and listed as follows.
First, the energy prices should be adjusted appropriately. The energy price effect was the biggest factor in mitigating the direct HCEs. It indicated that the deregulation of energy prices in the above text was an effective scientific way to realize energy conservation and emission reduction. The corresponding energy prices need to be further adjusted by government appropriately, especially the price of energy with a high emission coefficient. The six provinces should develop different policies for adjusting energy prices based on respective development characteristics, particularly those provinces that are greatly affected by fluctuations in energy price, such as Hunan Province, which needs to raise energy prices appropriately.
Additionally, the residents' consumption should be upgraded to a green and sustainable pattern. The effect of per capita consumption expenditure had the most impact on the growth of direct HCEs in Central China. With the development of economy and the improvement of residents' living standard, the government should advocate the shift of residents' consumption patterns to green and sustainable, and reasonably control the growth rate of transportation and energy-intensive devices. Some measures can be implemented. For example, the charging model of residential energy consumption needs to be adjusted, the energy-saving standards for household appliances need to be popularized, and the energy-saving behavior patterns need to be promoted and publicized with the aid of the media.
Next, population size and low-carbon urbanization should be reasonably controlled. The growth of direct HCEs in Central China has been less influenced by population size and population urban-rural structure. However, the net population continues to rise due to the large population base and the liberalization of the three-child policy in this region. Consequently, the government should control population growth within a reasonable range. This is particularly true for provinces with a large population, such as Henan Province. In addition, the urban pattern should be scientifically designed in the process of urbanization, and the development of low-carbon infrastructure should be encouraged. Meanwhile, raising people's awareness of energy conservation is an important goal in the cultivation of urbanized humanities. The urbanization can be developed along the new direction of intensive, green and low-carbon.
Lastly, the household energy structure should be optimized and the development and use of clean energy should be accelerated. Greater efforts should be made to optimize the household energy structure by reducing the use of coal and developing the use of natural gas, especially in rural areas, and encouraging the use of renewable energy. From the perspective of the emission structure of each province, the modification of Anhui's energy structure was most conducive to the reduction of CO 2 emissions. Shanxi Province needs to cut the use of fossil energy such as coal and oil, and accelerate the development of non-fossil fuel energy such as wind and nuclear, in comparison to other the five provinces.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.